Our objectives are: a) to test our computer algorithms for helping radiologists detect lung tumors in chest radiographs and measure the extent of ductal prominence in mammograms in a realistic clinical setting, b) to increase the sensitivity and accuracy of detecting radiographic images of lung tumors and measuring the extent of ductal prominence in mammograms at the UCI Medical Center and c) continued improvements in our computer algorithms for detecting, locating, enhancing, and analyzing medically important objects in radiographic images. Our research is organized into three units: 1) object detection and analysis, 2) image filtering, and 3) image-guided tomography. We shall test the clinical effectiveness of our systems for computer-aided detection of lung tumors in chest radiographs and for measuring the extent of ductal prominence in mammograms. We also plan to increase the power and speed of our computer techniques for aiding the analysis of radiographs: using heap sorts in boundary following, medically orientated data structures in digitized radiographs, textural features based on Markov models, and piecewise linear classifiers. We plan to exploit our techniques of block-approximated, zonal filtering for enhancement and deblurring of radiographs. We plan to assimilate our techniques of boundary detection into coarse-fine computed tomography, in order to obtain adequate detail in selected regions of interest with reduced doses.